Grain boundaries in polycrystalline metals are potent defect sinks. Alloying elements and impurity species will often readily segregate to boundaries, affecting the properties of metallic materials and in many cases increasing their susceptibility to failure. Thus there is a powerful need to understand the relationship between grain boundary geometry and impurity species type in determining segregation tendency and subsequent property change.
Unfortunately, whilst grain boundaries are extended defects, the process of segregation depends on atom level detail of the boundary structure and on the specific chemistry of the host and segregating elements. Density functional theory can handle the chemistry, but, due to computational expense, cannot address realistic systems containing general grain boundaries. Nevertheless, accurate predictions of segregation tendency are required as inputs for models at larger length scales and to support efforts in grain boundary engineering.
Here we present the results of high-throughput DFT simulations of grain boundary segregation in aluminium. We consider the grain boundaries as a collection of local environments, which in turn depend on the overall grain boundary geometry. We then consider how the interaction of a variety of segregating species depends on the local environment. This provides a link between boundary geometry and segregation behaviour that avoids the need for explicit DFT calculations of (perhaps impossibly) large systems.